U.S. patent number 4,408,344 [Application Number 06/252,554] was granted by the patent office on 1983-10-04 for ocr and bar code reader using multi port matrix array.
This patent grant is currently assigned to Recognition Equipment Incorporated. Invention is credited to Lynn D. McWaters, Medford D. Sanner.
United States Patent |
4,408,344 |
McWaters , et al. |
October 4, 1983 |
OCR and Bar code reader using multi port matrix array
Abstract
A common scanning unit is utilized in an optical character
reader which reads both alphanumeric and bar code characters. An N
column by M row area photosensor detects both the alphanumeric and
bar code signals. The system detects which type of character is
being read and then outputs the identity of the character.
Inventors: |
McWaters; Lynn D. (Garland,
TX), Sanner; Medford D. (Irving, TX) |
Assignee: |
Recognition Equipment
Incorporated (Irving, TX)
|
Family
ID: |
22956501 |
Appl.
No.: |
06/252,554 |
Filed: |
April 9, 1981 |
Current U.S.
Class: |
382/318; 235/436;
382/309; 382/324; 235/462.07; 235/440 |
Current CPC
Class: |
G06K
7/10881 (20130101); G06K 7/1092 (20130101); G06K
9/2009 (20130101); G06K 9/20 (20130101); G06K
9/228 (20130101) |
Current International
Class: |
G06K
7/10 (20060101); G06K 9/22 (20060101); G06K
009/28 () |
Field of
Search: |
;340/146.3Z,146.3D,146.3SY,146.3C ;235/440,454,462,463,470,472,494
;250/566,568-570,578 ;382/62,68,57 |
References Cited
[Referenced By]
U.S. Patent Documents
Primary Examiner: Boudreau; Leo H.
Attorney, Agent or Firm: Richards, Harris & Medlock
Claims
What is claimed:
1. An optical reader for reading alphanumeric and bar code
characters using a single area array of photo sensitive diodes
wherein there are at least two columns in the array and each column
has a plurality of photo sensitive diodes such that the column
exceeds in length the height of the character being read,
comprising; means for reading the electrical signals for each photo
diode in a column sequentially, means for reading the sequential
electrical signals for each column in parallel with the signals in
all the columns, means for selecting part of the signals from one
of the columns and summing the selected signals for one or more
scans to identify therefrom bar code characters being scanned,
means for utilizing electrical signals from all of the columns to
identify alphanumeric characters being scanned, and mdeans to
identify which type of character is being read, bar code or
alphanumeric, and out putting signals representative of the read
characters.
2. The reader according to claim 1, including means to combine said
part of the signals from one of the columns to produce a signal
equivalent to one which would be produced by an equivalent
rectangular photo diode.
Description
FIELD OF INNVENTION
This invention relates to optical character readers and more
particularly to a hand held optical character reader for reading
either alphanumeric or bar code format with the same hand held unit
with no operator intervention to switch between the reading of the
two formats.
PRIOR ART
Hand held optical readers are well known in the art and
alphanumeric readers are described in U.S. Pat. No. 4,075,605 and
4,188,687. There are also many patents relating to hand held units
for reading bar code particularly the universal product code (UPC),
however there are no hand held readers which read both bar code
format and alphanumeric format. With the increasing use of OCR
coding in merchandising tags and inventory control it is desirable
to have a hand held unit which will read either the bar code or the
alphanumeric format without the operator having to switch the unit
depending upon which code is to be read.
SUMMARY OF THE INVENTION
This invention relates to hand held optical readers which will read
either bar code or alphanumeric font automatically without any
switching by the operator. The data to be read is scanned with one
or more columnar arrays. The array may be a single column or a
multiple column. When the direction of reading is a criteria, then
at least two columns will be utilized. With at least two column the
array has parallel outputs which read out sequentially each of the
picture elements from the column but in parallel with all other
columns. The bar code signal is produced by selecting one of the
column outputs. Several picture elements are selected near the
center of the column. The output is from a column near the center
of the array in order to reduce the effects of rolling the hand
held reader and to obtain most optimum illumination. The bar code
signal is derived from only a few of the vertical pixels so a
reasonable amount of angular rotation of the sensor axis relative
to the vertical of the bar code can be tolerated.
DRAWINGS
FIG. 1 is a block diagram of the dual reading optical character
recognition system.
FIG. 2 illustrates both alphanumeric and bar code format.
FIG. 3 is a circuit diagram combining the functions of analog
gating summing of picture elements and sampling and hold.
FIG. 4 illustrates wave forms from the circuit of FIG. 3.
FIG. 5a is a block diagram of thresholding circuit.
FIG. 5b is a circuit diagram of a thresholding circuit.
FIG. 6 illustrates the wave forms of the Video inputs and outputs
of the comparator.
FIGS. 7a, 7b, and 7c are a microprocessor for reading the bar code
format.
FIGS. 8a-8c comprise a flowchart of the UPC barcode reading
process.
And tables 1, 2 and 3 illustrate the black/white decoding patterns
for UPC bar code format.
PREFERRED EMBODIMENT
Illustrated in FIG. 1 is preferred embodiment of the bar
code-alphanumeric reader. The scanning array is an N x M array
having one or more N columns each column having M elements. The
sequential output from each column is amplified and then applied to
a signal conditioning circuit which applies the video therefrom to
a character recognition unit. This character recognition unit may
be similar to that described in and illustrated in U.S. Pat. No.
4,075,605.
The analog signal for the bar code reader, processor 2, takes the
video out of only one column. In a milticolumn array the column
selected is near the center of the array; however, if only one or
two columns are used in a scanner then either the single or the
right or left column may be utilized. The bar code analog signal is
applied to analog switch which is under the control of the sample
logic. The switch outputs the amplitude of the specific picture
element within the self scanned sensor. This output constitutes the
bar code sensing site and can consist of a single pixel or a column
of several elements. Utilizing three or four picture elements will
effectively result in having a rectangular shaped bar code sensor
that will be tall and thin in the same orientation as the bars.
This characteristic will reduce the effects of voids in the bars.
The use of an analog switch to select specific picture elements
allows a low percentage of modulation signal to be accommodated
using low cost hardware. It is necessary to get accurate amplitude
samples in order to construct a bar code signal that will reflect
the correct bar widths.
The sample control logic samples signals from those picture
elements which are near the center of the column of the array being
used for reading bar codes.
From the analog switch the bar code signal is applied to the
summing circuit for sequential picture elements. The function of
the pixel summing circuit is to combine the outputs of the multiple
samples to produce a signal equivalent to that which would be
generated by an equivalent rectangular photosite. The cirucuit must
respond to sequential inputs and retain the average value until a
new set of samples is produced by the sensor.
A circuit which combines the function of analog gating, summing of
picture elements, and sample-and-hold is illustrated in FIG. 3. The
CA3080 circuit is an operational transconductance amplifier (OTA)
that can be gated on and off. The pixel sample pulse enables the
OTA which produces an output current proportional to the video
signal amplitude and polarity. This current charges capacitor C1
producing a voltage which is the integral of the input video
signal. Since the gating of the OTA is only during the desired
pixel times, the voltage on C1 will be proportional to the average
value of the pixels. This voltage level is buffered by the
Operational Amplifier to drive the low pass filter circuit.
Immediately before the next group of pixels are sampled, capacitor
C1 is discharge to zero.
The relative timing of the control pulses to operate the pixel
summing circuit is illustrated in FIG. 4.
The minimum bar width and space pair represents the highest spatial
information frequency. When this pattern is scanned by a sensor a
signal frequency is obtained that will be in the range of 1500 Hz
for a 30 IPS scan rate. The low pass filter is designed to pass
data only up to this frequency; thus improving the signal-to-noise
ratio of the signal before it is thresholded. The filter utilized
for the low pass function is an active filter which implements a
3-pole, low ripple Chebyshev design.
The function of the thresholding circuit is to transform the analog
signal that is derived from optically scanning the bar code into a
digital signal level whose pulse widths are representative of the
widths of the bars in the bar code. The threshold circuit must
accommodate a wide range of signal levels and modulation
percentages in order to read a high percentage of tags or inventory
codes. Because of this, an adaptive reference level must be used in
thresholding the analog signal. FIG. 5a shows a block diagram of
this approach. The circuit to accomplish this function is shown in
FIG. 5b. The dynamic reference voltage is either 0.6 volts less
than a positive peak or 0.6 volts greater than the negative peak.
The peak values of the analog signal establish new reference levels
for each signal swing and a black-to-white or white-to-black output
transition occurs when the analog signal decreases from its peak
value by 0.6 volts. Pulse widths will be accurate for waveforms
that have equal positive and negative slopes and have peak to peak
swings of 1.2 volts or greater. The waveforms shown in FIG. 6
represent a very poor analog input, but show the reference that
would be produced and the resulting digital output. The output
pulse widths are equal to the time between analog peaks or, in the
case of a flat top signal, the time between positive to negative
(or vice versa) directional changes. The comparator (FIG. 5a) is
biased by the quiescent reference to produce a white output so that
as the reader is brought closer to the paper no change in output
occurs.
The character detection logic is implemented in the form of a 3874
8-bit microprocessor and is illustrated in FIG. 7. The
microprocessor looks for black/white transitions on its input port
5-bit 7. The time intervals between black white transitions are
then recorded in the microprocessor's memory. When the time between
the black white transitions is long (i.e. widest bar width divided
by minimum reading speed), the microprocessor then "times out" and
goes to process the black white transitions for a valid bar code
value.
In the case of the UPC code, each set of two black bars and two
white bars represents digital character. The relative widths of the
black and white determine which digital characters. The Tables 1, 2
and 3 illustrate the decoding for UPC characters.
The velocity variations of a hand held device can be compensated
for with this bar code due to the fact that each set of two black
bars and two white bars compose 7 modules.
For:
Tw=Time for white
Tb=Time for black
Tm=Time for 1/7 module
Then:
W.sub.w1 =T.sub.w1 /Tm
W.sub.b1 =T.sub.b1 /Tm
W.sub.w2 =T.sub.w2 /Tm
W.sub.b2 =T.sub.b2 /Tm
And:
W.sub.w1 +W.sub.w2 +W.sub.b1 +W.sub.b2 =7
W.sub.min =1
W.sub.max =4
These characteristics can then be used to decode the black white
transitions into decimal characters.
A flow chart of the UPC barcode reading process is shown in FIG.
8.
Step 1. The circuits and registers are initialized by power on
clear circuits at initial power turn on. The time for the black and
white bars and spaces is recorded in memory during Step 2. If a
period of time has passed which is greater than the widest bar or
space divided by the lowest expected reader velocity, and no black
to white or white to black data transition has occurred, then a
time out condition is generated at Step 3. If fewer than 33
transitions of black to white and white to black have occurred at
Step 4, then it is impossible for a full UPC barcode to have been
read. The process is reinitialized to Step 1.
With 33 or more transitions of black to white and white to black it
is possible that a valid UPC barcode has been scanned and the
process advances to Step 5 where an initial guess of a right to
left scan of the reader over the barcode is assumed.
At Step 6 the collected data from Step 2 is looked at as groups of
2 bars and two spaces since all valid numbers are so grouped as
shown in Table 1. If the number of transitions is greater than 55,
then the process assumes a twelve digit long code version the UPC
barcode at Step 7, and proceeds to Step 8 where the mask for the
four bars is derived by using the equations:
W.sub.W1 +W.sub.W2 +W.sub.B1 +W.sub.b2 =7 and
W.sub.min =1
W.sub.max =4
The sum of the widths of the two black (W.sub.bx) bars and two
white spaces (W.sub.wx) is 7 units and the minimum width is one and
the maximum width is four.
In Step 9 the character mask or pattern is checked against the odd
parity characters of Table 3. If a valid character is found, then
the decision goes to Step 10 where the first valid odd parity
character is compared to the numbers one (1), seven (7) and eight
(8). These numbers are excluded since in the 12 digit UPC, the
first left hand digit cannot be one, seven or eight. The negative
leg of Step 9 and the positive leg of Step 10 will be discussed
later as they indicate a good start character for the 12 digit UPC
code has not been derived yet.
Step 11 indicates that the first valid left hand odd parity
character of the 12 digit UPC code has been found and its place in
the input data stream is marked and the next two sets of black and
white bars are indexed for the character mask derivation in Step
12. This is the same process as described in Step 8. If this
character mask corresponds to a valid odd parity character in Step
13, then this character is stored away and the input stream
reindexed to next two sets of black and white bars. If less than
six characters have been found, then Steps 12 and 13 are repeated
by the decision at Step 14.
After 6 characters have been found by repeating Steps 12 and 13,
the bar code format expects a center bar pattern as shown in Table
1. Step 15 skips over this center bar pattern. Step 16 looks at two
more sets of bars and spaces as was done in Step 8. Since the right
hand side of the 12 digit UPC code is even parity, Step 17 looks
for valid even parity digits from the complement of the odd parity
of Table 3. This process continues until six even parity digits are
found, thus making a total of 12 character as indicated in Step 18.
When the 12th character is found, a check digit is calculated over
the 12 characters in Step 19, and if the check digit calculates
properly the right guard bars are validated in Step 20 against the
pattern shown in Table 1.
When all of these checks, as outlined above, are passed, the digits
for the validated bar code data are output in Step 21 to the output
selector and interfaceed as shown in FIG. 1a. The steps as outlined
above for Steps 8 through 21 was for a 12 digit UPC barcode reading
the data out from left to right as it appears on the page. The
actual scanned direction was compensated for in Steps 5 or 35.
Now returning to Step 7 and taking the "no leg", this part of the
flow chart is basically for the E version (6 digit) UPC code. At
Step 22 if the number of transitions (bars and spaces) is less than
29, then this is insufficient for a valid 6 digit barcode and the
process goes to Step 34 to try the opposite direction, if it has
not already been tried. At Step 23 the number of transitions is
checked to be less than 44. If this test is passed then Step 24
uses a set of two bars and two spaces to derive a character mask as
was done in Step 8. Since at the present time only number system
zero from Table 2 is used, then this first character must be even
parity. This check is done in Step 25.
If a valid first left character for the 6 digit UPC barcode is
found, the input pointer is backed up three transitions and the
left guard bar is checked for the pattern shown in Table 1. This is
Step 26 of the process. If the left guard bar is validated then the
location of the first valid character is recorded in Step 27. The
pointer is moved to the next set of two black and two white bars
and the character mask is derived in Step 28 as it was in Step
8.
In Step 29 the character mask is compared to Table 3 for a match of
the odd or even parity digits. If a match is found, not a reject,
the character and parity is stored and the process of Step 28 and
29 is repeated until six characters are found in Step 30. When the
six characters have been found, Step 31 compares the check digit
with the character parity shown in Table 2. If this check digit
calculation passes, then in Step 32, the right guard bars, as shown
in Table 1, are validated. If this validation passes then the data
is output in Step 33 as it was in Step 21. This thus completes the
successful decode of a six character UPC barcode.
Returning to Step 34 which is entered from Steps 20, 22 or 23.
These indicate a failure of the decode process in the initial
assumed negative direction. This in Step 34 if the present
direction is negative, then the positive direction is assumed in
Step 35 and the process restarts at Step 6 for another look in the
opposite direction. If, in Step 34, the previous direction was
positive, that implies Step 35 has already been tried before so the
process aborts having not found a valid barcode in either
direction.
Step 36 is a result of a failure in Steps 9, 10, 25, 26 or 37. In
this case if the direction was positive or negative the pointer is
moved one bar or space in the previously defined direction and the
process tried over again at Step 6.
Step 37 is a result of a failure in Steps 13, 29, 31 or 32. In this
case the pointer to the input data stream must be replaced to its
previous origin before the one bar or space adjustment is made in
the proper direction.
The UPC bar code is further protected with stop/start bars, center
bars, right and left of center bars and overall parity check
characters.
These are generally illustrated in table 1. Other black/white bar
codes have similar properties and can be decoded by recording the
time length of the black/white bars and using appropriate decoding
equations.
The line edit logic is performed by a 3874 Microprocessor. The
digits as determined in the character detection logic are checked
as to direction of scan, digit parity, and line parity. In the case
of the UPC bar code, the number of black elements (not bars) per
digit determine the character parity (000 is left hand bars; even
is right hand bars). The line parity is determined by a weighting
scheme of the first eleven characters to determine the twelfth
character.
These checks, coupled with right hand, center, and left hand guard
bars are used to provide bar code line data integrity.
When the bar code field has passed all of the above tests for data
integrity, the data is transferred to the output selector and the
interface.
Processor 1 may be similar to that described in U.S. Pat. Nos.
4,075,605 and 4,118,687 and is not further described here. The
output selecting an interface which is fed by both the alphanumeric
and the bar code processors gives priority to an output from
processor 1 if there is no output in processor 1 and there is
output in processor 2 then the character data output data is from
processor 2. In practice there would only be an output from both
processors if the units scanned as illustrated FIG. 2 covering the
alphanumeric characters and the bar code otherwise, there would
only be an output from one of the processors and the decision does
not have to be made.
While specific examples have been given of a combination bar
code/alphanumeric reader which will read either of the formats
without interference from the operator other examples will be
apparent to those skilled in the art having seen these specific
examples and the attached claims.
* * * * *